• The Journal of the Acoustical Society of Korea
• /
• v.19 no.1
• /
• pp.11-15
• /
• 2000

Speaker adaptive continuous speech recognition for the RM speech corpus is described in this paper. Learning of hidden markov models for the reference speaker is performed for the training data of RM corpus. For the evaluation, evaluation data of RM corpus are used. Parts of another training data of RM corpus are used for the speaker adaptation. After dynamic time warping of another speaker's data for the reference data is accomplished, error back propagation neural network is used to transform the spectrum between speakers to be recognized and reference speaker. Experimental results to get the best adaptation by tuning the neural network are described. The recognition ratio after adaptation is substantially increased 2.1 times for the word recognition and 4.7 times for the word accuracy for the best.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.13 no.6
• /
• pp.487-491
• /
• 2008

Tuning of the speed controller in the linear servo applications needs the accurate information of a mover mass including a load mass. Therefore this paper proposes the mass estimation method of a permanent magnet linear synchronous motor(PMLSM) applied at the vertical axis by using the recursive Least-Squares estimation algorithm. First, this paper derives the deterministic autoregressive moving average(DARMA) model of the mechanical dynamic system used at the vertical axis. The application of the Least-Squares algorithm to the derived DARMA model gives the mass estimation method. Matlab/Simulink-based simulation and experimental results show that the total mover mass of a PMLSM applied at the vertical axis can be accurately estimated at both no-load and load conditions.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.2
• /
• pp.837-856
• /
• 2016

Underwater image enhancement has received considerable attention in last decades, due to the nature of poor visibility and low contrast of underwater images. In this paper, we propose a new automatic underwater image enhancement algorithm, which combines nonsubsampled contourlet transform (NSCT) domain enhancement techniques with the mechanism of the human visual system (HVS). We apply the multiscale retinex algorithm based on the HVS into NSCT domain in order to eliminate the non-uniform illumination, and adopt the threshold denoising technique to suppress underwater noise. Our proposed algorithm incorporates the luminance masking and contrast masking characteristics of the HVS into NSCT domain to yield the new HVS-based NSCT. Moreover, we define two nonlinear mapping functions. The first one is used to manipulate the HVS-based NSCT contrast coefficients to enhance the edges. The second one is a gain function which modifies the lowpass subband coefficients to adjust the global dynamic range. As a result, our algorithm can achieve contrast enhancement, image denoising and edge sharpening automatically and simultaneously. Experimental results illustrate that our proposed algorithm has better enhancement performance than state-of-the-art algorithms both in subjective evaluation and quantitative assessment. In addition, our algorithm can automatically achieve underwater image enhancement without any parameter tuning.

• Proceedings of the KIEE Conference
• /
• 2001.07a
• /
• pp.134-136
• /
• 2001

This paper describes the application of Fuzzy Power System Stabilizer(FPSS) for improving dynamic stability of power system. The Real-coding Genetic Algorithm(RGA) was applied to optimize gains of the inputs and outputs of the FPSS. The effectiveness of the proposed FPSS was demonstrated by simulation studies for single-machine infinite system. To show the superiority of the proposed FPSS, its performances were compared with those of Conventional Power System Stabilizer(CPSS) The proposed FPSS showed better control performances than the CPSS in three-phase ground fault under a normal load which was system condition in tuning FPSS. To show the robustness of the proposed FPSS, it was applied to damp the low frequency oscillations caused by disturbances such as three-phase ground fault under heavy and light load conditions. The proposed FPSS showed better performance than CPSS in terms of the settling time and damping effect for power system operation condition.

• Journal of Power Electronics
• /
• v.14 no.5
• /
• pp.1008-1027
• /
• 2014

An electric scooter with a V-belt continuously variable transmission (CVT) driven by a permanent magnet synchronous motor (PMSM) has a lot of nonlinear and time-varying characteristics, and accurate dynamic models are difficult to establish for linear controller designs. A PMSM servo-drive electric scooter controlled by a novel hybrid modified recurrent Legendre neural network (NN) control system is proposed to solve difficulties of linear controllers under the occurrence of nonlinear load disturbances and parameters variations. Firstly, the system structure of a V-belt CVT driven electric scooter using a PMSM servo drive is established. Secondly, the novel hybrid modified recurrent Legendre NN control system, which consists of an inspector control, a modified recurrent Legendre NN control with an adaptation law, and a recouped control with an estimation law, is proposed to improve its performance. Moreover, the on-line parameter tuning method of the modified recurrent Legendre NN is derived according to the Lyapunov stability theorem and the gradient descent method. Furthermore, two optimal learning rates for the modified recurrent Legendre NN are derived to speed up the parameter convergence. Finally, comparative studies are carried out to show the effectiveness of the proposed control scheme through experimental results.

• Structural Engineering and Mechanics
• /
• v.36 no.5
• /
• pp.561-573
• /
• 2010

The potential of the liquid column vibration absorber (LCVA) as a seismic vibration control device for structures has been explored in this paper. In this work, the structure has been modeled as a linear, viscously damped single-degree-of-freedom (SDOF) system. The governing differential equations of motion for the damper liquid and for the coupled structure-LCVA system have been derived from dynamic equilibrium. The nonlinear orifice damping in the LCVA has been linearized by a stochastic equivalent linearization technique. A transfer function formulation for the structure-LCVA system has been presented. The design parameters of the LCVA have been identified and by applying the transfer function formulation the optimum combination of these parameters has been determined to obtain the most efficient control performance of the LCVA in terms of the reduction in the root-mean-square (r.m.s.) displacement response of the structure. The study has been carried out for an example structure subjected to base input characterized by a white noise power spectral density function (PSDF). The sensitivity of the performance of the LCVA to the coefficient of head loss and to the tuning ratio have also been examined and compared with that of the liquid column damper (LCD). Finally, a simulation study has been carried out with a recorded accelerogram, to demonstrate the effectiveness of the LCVA.

• Proceedings of the KIEE Conference
• /
• 1996.07b
• /
• pp.1359-1361
• /
• 1996

This paper describes a design of real time fuzzy controller using Minimum fuzzy control Rule Selection Method(MRSM). The control algorithm of dynamic systems needs less computation time and memory. To reduce the computation time of fuzzy logic controller, minimum number of rules are to be selected for the fuzzy input variable. The universe of discourse is divided by the number of linguistic labels to allocate the assigned membership function to the fuzzy input variables. In this case, since fuzzy input variables are continuous, scale factor SU is tuned independently. According to increment of SU control surface is improved to adapt the change of system parameter. At this, crisp control surface is increased. With the increament of crisp control surface, fuzzy control surface is reduced. When error state deviates from desirable error state, crisp control surface is more useful than fuzzy control surface for obtaining fast rising time.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.6 no.5
• /
• pp.432-437
• /
• 2001

This paper describes a new developed 3.3KV/4MW class three-level Voltage Source Inverter(VSI), which is equipped with IIMS(Inverter Information Management System) based on the world wide web and with the Virtual operation simulator. The algorithm for motor control is the stator oriented Direct Torque Control(DTC), which works without speed sensor and gives the physically fastest dynamic response. The IIMS have the functions of operation monitoring and data managements. Virtual operation simulator can analyze and tune the system characteristics without main power. Now, this system is under the field test to verify the confidence.

• Nuclear Engineering and Technology
• /
• v.53 no.7
• /
• pp.2251-2260
• /
• 2021

The motion control of the divertor maintenance system of the China Fusion Engineering Test Reactor (CFETR) was studied in this paper, in which CFETR Multi-Functional Maintenance Platform (MFMP) was simplified as a parallel robot for the convenience of theoretical analysis. In order to design the motion controller of parallel robot, the kinematics analysis of parallel robot was carried out. After that, the dynamic modeling of the hydraulic system was built. As the large variation of heavy payload on MFMP and highly nonlinearity of the system, A Fuzzy-PID controller was built for self-tuning PID controller parameters by using Fuzzy system to achieve better performance. In order to test the feasibility of the Fuzzy-PID controller, the simulation model of the system was built in Simulink. The results have showed that Fuzzy-PID controller can significantly reduce the angular error of the moving platform and provide the stable motion for transferring the divertor.

• The Journal of the Acoustical Society of Korea
• /
• v.37 no.6
• /
• pp.475-482
• /
• 2018

In this paper, we present a study on data augmentation methods for DNN (Deep Neural Network)-based audio tagging. In this system, an audio signal is converted into a mel-spectrogram and used as an input to the DNN for audio tagging. To cope with the problem associated with a small number of training data, we augment the training samples using time stretching, pitch shifting, dynamic range compression, and block mixing. In this paper, we derive optimal parameters and combinations for the augmentation methods through audio tagging simulations.